Platform Overview

CSCS offers cscs2go as a means to lower barriers and simplify access to world-class HPC and storage services, without having to go through a peer-review process, via the so-called HPC Platform.

Thanks to cscs2go users can move their data, begin their analyses and run their simulations within a few days of the request, with little overhead. Upon registration, a user will be granted a project with disk space for data storage and a quota of node-hours to consume within one year on a CSCS HPC resource. That’s it!

Once a project is enabled, users gain access to the HPC Platform with a variety of supported services, including:

  • Compute: Users get a node-hours allocation on multicore or GPUs to use via a queuing system. Jobs can be submitted using the user's own compiled codes or codes provided by CSCS. I/O operations must utilize the "Scratch" file system, requiring data staging before and after job execution (e.g., utilizing the Data Mover service).
  • Storage: Users get limited space for configuration data and codes in a "Home" file system, and a larger quota in a file system for scientific data and results. Both file systems are automatically backed-up on a daily basis and data can be transferred to and from any remote site using the Data Mover service.
  • Data Science and Machine Learning: A selection of applications is available, leveraging the state-of-the-art HPC environment for data science and machine learning tasks.
  • Scientific Applications: Users have access to a range of scientific applications spanning from quantum physics to supramolecular dynamics simulations.
  • Programming environment: A selection of compilers, debuggers and performance tools to help you get the best from your code.
  • Interactive Supercumputing: A selection of interactive languages that can be executed on the supercomputer via a JupyterLab service.

The updated list of services available at CSCS can be found here.

All these services are supported by our Help Desk. For those who already know CSCS, cscs2go shares many features with the User Lab.